oracle

What problem does it solve?

Oracle solves the critical problem of context loss and repetitive mistakes in AI-assisted development. It prevents valuable learnings from being forgotten across sessions, eliminates the need to re-explain project specifics, and reduces wasted tokens by providing intelligent, relevant context. This accelerates development, improves code quality, and ensures AI works efficiently for you.

Core Features & Use Cases

• Intelligent Context Injection: Automatically loads project-specific patterns, preferences, gotchas, and corrections at session start or for specific tasks, ensuring Claude is always context-aware and productive.
• Learning from Corrections: When you correct Claude, Oracle records the mistake and the right approach, preventing recurrence and continuously improving AI performance and accuracy.
• Session Recording & Timeline: Tracks all interactions, changes, decisions, and learnings, building a comprehensive, searchable project history that serves as institutional knowledge.
• Automation Opportunities: Analyzes session history to detect repeated tasks and suggests or generates scripts for automation, saving time and tokens on deterministic operations.
• Use Case: A developer frequently makes a specific error in React state management. Oracle, after being corrected once, records this as a "correction" with high priority. In subsequent sessions, when the developer works on React components, Oracle proactively injects this correction into the context, reminding Claude (and the developer) of the best practice, thus preventing the mistake from happening again and saving valuable debugging time.

Quick Start

To initialize Oracle for your project: python .claude/skills/oracle/Scripts/init_oracle.py To record what you learned after a session: python .claude/skills/oracle/Scripts/record_session.py --interactive To load relevant context for your next session: python .claude/skills/oracle/Scripts/load_context.py